Predicting extreme rainfall in regional areas of Bangladesh: A Bayesian approach

Hossain, Moyazzem; Fawcett, Lee; Golightly, Andrew

Abstract

Extreme weather events are anticipated to become more common around the world and they impact yield volatility i.e. reduce food production. Given the changing nature of the world s climate, and the disproportionate effect this might have on developing countries such as Bangladesh, this is an important topic to study. Agriculture is the major employment source and a significant economic contributing sector in Bangladesh. Moreover, extreme rainfall has a significant effect on agricultural production, which negatively affects the nation's food security and may make it more difficult to end hunger and achieve United Nations Sustainable Development Goal 2. Therefore, understanding and modelling the extremes of rainfall in Bangladesh is crucial. This study considers extreme rainfall in different regional domains of Bangladesh and estimates predictive return levels using the Generalized Extreme Value (GEV) and Generalized Pareto Distribution (GPD) in the Bayesian setting. Finally, a comparative study is carried out among return levels at regional areas determined by the distributions considered here. Results depict that in the case of the GEV, once every 100 years, on average, we can expect daily rainfall levels to exceed 400 mm in some locations. However, in the case of the GPD, once every 50 years, on average, we can expect daily rainfall levels to exceed 800 mm in Dinajpur and Mymensingh regions. More rainfall will be observed in Chattogram, Cox s Bazar, Dinajpur, Faridpur, Khulna, and Mymensingh regions compared to other parts of Bangladesh. It is also observed that the 100-year return levels are closer to the lower bound than the upper bound of the credible intervals. This information may also be used to identify regions that are particularly vulnerable to the kind of heavy rain that causes flooding.

Citation

Hossain, M., Fawcett, L., & Golightly, A. (2023). Predicting extreme rainfall in regional areas of Bangladesh: A Bayesian approach. Australasian Journal of Regional Studies, 29(1), 114-132

• Journal Article Type: Article
• Acceptance Date: Oct 26, 2023
• Online Publication Date: Jan 1, 2023
• Publication Date: 2023-11
• Deposit Date: Feb 23, 2025
• Journal: Australasian Journal of Regional Studies
• Print ISSN: 1324-0935
• Peer Reviewed: Peer Reviewed
• Volume: 29
• Issue: 1
• Pages: 114-132
• Public URL: https://durham-repository.worktribe.com/output/3494775